2
$\begingroup$

Let $\psi_1$ be a Normal random variable with mean $\mu_1$ and standard deviation $\sigma_1$. Let $\xi$ be defined as

$$ \xi=c\,\mathbb{1}_{\left\{\psi_2+\psi_1\leq 0\right\}}, $$ where $\mathbb{1}$ is the indicator function, and $\psi_2$ a Normal random variable with mean $\mu_2$ and standard deviation $\sigma_2$. Thus $\xi$ is a discrete random variable that can be either $c$ or $0$. The problem is to compute the following CDF:

$$ F\left(\alpha\right)=\mathbb{P}\left[\psi_2+\xi\leq \alpha\right]. $$

Since the variable $\xi$ is discrete I cannot use

$$ \mathbb{P}\left[X+Y\leq \alpha\right] = \int_{-\infty}^{\infty}\int_{-\infty}^{v=\alpha-u}f_{X,Y}\left(u,v\right)\,du\,dv = \int_{-\infty}^{\infty}\int_{-\infty}^{v=\alpha-u} f_{Y\mid X}\left(v\mid u\right)\,f_{X}\left(u\right)\,dv\,du $$

$\endgroup$
2
$\begingroup$

Recall that $\{ \psi_2 + \xi \leqslant \alpha \} = \{\psi_2 + c [ \psi_2+\psi_1 \leqslant 0] \leqslant \alpha \} = \{\psi_1 \leqslant -\psi_2, \psi_2 + c \leqslant \alpha \} \lor \{\psi_1 > -\psi_2, \psi_2 \leqslant \alpha \} $ and the latter two events are disjoint. Hence $$ \begin{eqnarray} F(\alpha) &=& \Pr\left(\psi_2 + \xi \leqslant \alpha\right) = \Pr\left(\psi_1 \leqslant -\psi_2, \psi_2 + c \leqslant \alpha\right) + \Pr\left(\psi_1 > -\psi_2, \psi_2 \leqslant \alpha\right) \\ &=& \Pr\left( \psi_2 \leqslant \min(\alpha-c, -\psi_1) \right) + \Pr\left(-\psi_1 < \psi_2 \leqslant \alpha\right) \\ &=& \mathbb{E}\left( \Phi\left(\frac{\min(\alpha-c, -\psi_1)-\mu_2}{\sigma_2}\right) \right) + \mathbb{E}\left( \Phi\left( \frac{\alpha-\mu_2}{\sigma_2}\right) - \Phi\left( \frac{-\psi_1-\mu_2}{\sigma_2} \right)\right) \\ &=& \Phi\left( \frac{\alpha-\mu_2}{\sigma_2}\right) + \mathbb{E}\left( \Phi\left(\frac{\min(\alpha-c, -\psi_1)-\mu_2}{\sigma_2}\right) - \Phi\left( \frac{-\psi_1-\mu_2}{\sigma_2} \right) \right) \\ &=& \Phi\left( \frac{\alpha-\mu_2}{\sigma_2}\right) + \mathbb{E}\left( \Phi\left(\frac{\min(\alpha-c, -\psi_1)-\mu_2}{\sigma_2}\right) - \Phi\left( \frac{-\psi_1-\mu_2}{\sigma_2} \right) \mid \alpha-c < -\psi_1 \right) \\ &=& \Phi\left( \frac{\alpha-\mu_2}{\sigma_2}\right) + \mathbb{E}\left( \Phi\left(\frac{\alpha-c-\mu_2}{\sigma_2}\right) - \Phi\left( \frac{-\psi_1-\mu_2}{\sigma_2} \right) \mid \alpha-c < -\psi_1 \right) \\ &=& \Phi\left( \frac{\alpha-\mu_2}{\sigma_2}\right) + \Phi\left(\frac{\alpha-c-\mu_2}{\sigma_2}\right) \Pr\left(\psi_1 < c-\alpha\right) - \mathbb{E}\left(\Phi\left( \frac{-\psi_1-\mu_2}{\sigma_2} \right) \mid \alpha-c < -\psi_1 \right) \\ &=& \Phi\left( \frac{\alpha-\mu_2}{\sigma_2}\right) + \Phi\left(\frac{\alpha-c-\mu_2}{\sigma_2}\right) \Phi\left(\frac{ c-\alpha - \mu_1}{\sigma_1}\right) - \mathbb{E}\left(\Phi\left( \frac{-\psi_1-\mu_2}{\sigma_2} \right) \mid \alpha-c < -\psi_1 \right) \end{eqnarray} $$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.